There are numerous options to try, and the result will be different each time, so you will have to experiment with your own data. As for maximum sets to combine for each planet, it's not so much the maximum number, but the total duration in time. For Jupiter, I take 3 minute videos and can easily combine 20 minute worth of integrations. For Saturn and Mars, I take 6 minute individual videos, and can combine 30 minutes of data in Winjupos without problem.
With regards to processing, I always process each individual video, complete with wavelet sharpening, before combining in Winjupos. But it is often best to undershapen just slightly, so that you leave room for extra sharpening at the end. However, I think of Winjupos as more of a noise reduction tool (or an SNR booster, if you prefer), and typically don't do that much sharpening afterwards. You can easily overdue it if you are expecting to do significant sharpening after combining. The real question, is whether the derotated images are better or worse than the individual captures. What you will often find, is that if you take 10 videos of a planet in rapid succession, not all videos give the same quality, because of slight differences in the seeing conditions during each video. To asses this, you need to process and sharpen them individually. Usually, out of 10 videos, maybe 3-4 yield a significantly sharper result than the others. It makes sense to only combine these sharper results in Winjupos. Otherwise, the derotated image is smoother, but less detailed than the individual captures. And sometimes, no matter what you do, Winjupos does not give a better result than taking the single best capture and just working on that file. You never know until you test each set of data. Obviously, this experimentation can take quite some time, so it's probably only worth doing on nights of really good seeing where the data is worthwhile. If the results of each individual capture are only average quality, Winjupos isn't going to help much.
There's really no set limit on how many individual images you can combine, but if the total integration period gets very long (more than 30 minutes or so), this can lead to less than optimal results. But as always, trial and error rules the day. Once you have made measurement files for each image, it is very easy to change the number of images that you combine.